Make it from a playbook.
- In Playbooks and Schedules, select the verified script and press Launch.
- Select nodes, groups or tags and get a preview.
- Version, steps, and a complete list of goals.
- In the last step, select on schedule.
- Specify the time zone, hour, minute and days of the week.
- Choose what to do with the lost time and save it.
Asia/Yekaterinburg, or UTC. Time refers to the selected zone, not necessarily to the time zone of the server or browser.
Although the interface may call launches cron schedules, the form specifies the time and days of the week, and the five-field arbitrary expression is not entered into it.
An example of regular diagnosis
Create a playbook with Diagnostics host, check it manually on selected nodes. Then set 09:00,Asia/Yekaterinburg, Monday-Friday and Skip at skip.
In the list, check Next launch of, number of targets and status. After the time comes, open the schedule history, then the task created and the results of the nodes. The diagnostic snapshot shows the state at the time of execution, not continuous monitoring.
Scheduling changes
Routinely running a script can change servers in the same way as manual. Confirmation refers to a specific scenario and goals; check for repeatability and recovery before creating. For typed modifying actions, such as restarting a service or applying a file, the server requires explicit permission to change and rollout with a canary. Advanced rollout configuration is done via the preview API; the current wizard does not have these fields. If such a scenario is rejected, this restriction cannot be circumvented by pressing a button again. Use the correct API preview or manual launch. Script-only scripts support independent targets without the need for a canary, so a normal scripting schedule should not be considered a protected automatic stoppage of the entire park after a first error. task.Passes and intersections
On the day of the clock transfer, the missing local time is skipped, the repeating hour does not create two normal starts in one calendar day.
Service window through API
The schedule API further supportsmaintenance_window and the ends_at deadline. The window is set by local time and must contain the assigned moment; the end before the start means the transition through midnight.
A catch-up run is only possible while the corresponding window is open. Upon closing, the event receives a skipped_window. The window restricts the output of further work, but is not a general automatic rollback of actions already performed.
These settings should not be searched for in the current basic form: they refer to the advanced configuration of the /api/v1/schedules API.
